COMPACTION VOIDS FORMATION ON CONCRETE SURFACES: INFLUENCE OF RHEOLOGICAL PROPERTIES

Transcription

1 COMPACTION VOIDS FORMATION ON CONCRETE SURFACES: INFLUENCE OF RHEOLOGICAL PROPERTIES M. Martin (1,2), R. Pleau (1) and J. L. Gallias (2) (1) CRIB, Laval University, Canada (2) Laboratoire Matériaux et Sciences des Constructions, Cergy-Pontoise University, France Abstract Concrete surfaces may present surface defects, such as compaction voids. Mechanisms and parameters conditioning the concrete surface s texture have been little investigated in the scientific literature. This experimental study aims to assess the influence of the paste volume, the water/cement ratio (W/C), and the rheological properties of concrete on the compaction voids. Five series of concrete mixtures were tested in order to study the formation and the behaviour of compaction voids. All the concrete mixtures were based on an identical optimized granular skeleton made of 50% of sand and 50% of coarse aggregate. In the first two series of mixtures, all the parameters were kept constant except the paste volume and the W/C. In the third and fourth series of mixtures, all the parameters were kept constant except the dosage of the superplasticizer. In the fifth series, all the parameters were kept constant except the dosage of the viscosity-modifying agent. The rheological properties and surface area of compaction voids were measured for all concrete mixtures. The results obtained show that the amount of compaction voids decreases with an increase of the mortar volume. They also point out the existence of a threshold value of slump and yield stress below or beyond which, the amount of compaction voids becomes significant. 1. INTRODUCTION Since a few years, the requirements of architects and building owners as regards with the quality of the concrete surfaces become more stringent. In many cases, concrete surfaces are expected to be practically free of compaction voids or other visible defects. Those aesthetic requirements are no longer limited to architectural works, but tend to spread over a wide variety of construction works. Aesthetic quality of concrete surface is characterized by its colour and its texture. The colour of concrete surface depends on the chemical composition of cement and aggregates, as well as the porosity of the concrete skin (e.f. the thin layer of cement paste just beneath the surface). Casting conditions also have a significant influence. The development of tools to characterize the colour of concrete surfaces, and the study of the various parameters which influence this colour, were recently investigated [1,2]. Those studies propose the first Page 1

2 scientific approach to investigate the colour as a measurable aesthetic property of concrete, and they complement a number of qualitative studies carried out before [3,4]. Beyond colour, the aesthetic quality of concrete surfaces is also related to their texture (in terms of roughness and brightness), which depends on the chemical and physical properties of the concrete skin at a microscopic level. However, the quality of concrete surfaces is often detrimentally influenced by the presence of voids, cracks or other defects. A few studies on the formation of compaction voids are reported in the literature [5,6], and a method was recently developed in order to measure the void-size distribution at the surface of concrete using image analysis techniques [2]. But the parameters influencing the presence of compaction voids (also called blisters or bug holes), as well as the mechanisms governing their formation, were very little investigated even though those voids are the most common type of defects on concrete surfaces. Notwithstanding the aesthetic considerations, the absence of voids also offers many advantages:! it makes concrete more competitive as compared with other building materials;! it reduces the need of surface treatments prior the application of thin coatings or paintings on concrete;! it contributes to protect steel rebars against corrosion;! it reduces the costs associated with surface repair work. Most of the time, the compaction voids are irregular in shape and their mean diameter ranges from 1 to 10 mm. They result from a lack of compaction of concrete during casting. The presence of voids is related to the ability of the cement paste and/or the mortar to flow through granular skeleton in order to fill the empty spaces left at the surface of formwork. The volume of these empty spaces is conditioned by the compactness of the granular skeleton and the relative importance of the wall-effect, which disturb the stacking of particles near formwork. Qualitative studies [6,7] showed that the aggregate grading of the coarse aggregate, the shape of the particles, the fineness modulus of sand, the volume and viscosity of cement paste have a significant influence on the formation of the surface voids. 2. EXPERIMENTAL PROGRAM Our experimental study has for objective to assesss the influence of the paste volume, water/cement ratio (W/C), and rheological properties of concrete on the characteristics of the compaction voids at the surface of concrete. Five series of concrete mixtures made of the same optimized granular skeleton (with 50% of sand and 50% of coarse aggregate) were produced to take into account all the parameters involves. In the first two series, all the parameters were kept constant except the paste volume and the W/C. In the third and fourth series, the dosage of superplasticizer was the only variable involved. In the fifth series, a viscosity modifying agent was used at different dosages for mixtures having the same composition. All the concrete mixtures were batched in a vertical-axis mixer according to the following sequence:! the dry ingredients were first mixed for 2 minutes to reach homogeneity;! the water was the added and the mixing prolonged for 1 minute;! when needed, the superplasticizer was added and mixing prolonged for 1 minute;! when needed, the welan gum was added and mixing prolonged for 1 minute;! the concrete was mixed for an additional 2 minutes. Page 2

3 After mixing, the slump was measured as well as the rheological properties, and the concrete mixtures were cast in four vertical moulds of 10x15x20 cm in two layers consolidated using a vibrating needle. Moulds were made of plexiglass plates covered with a thin layer of emulsion to prevent bonding. The concrete specimens were demoulded 24 hours after mixing and stored in the laboratory until testing. 2.1 Materials and mixture composition All the concrete mixtures were made using an ASTM Type I ordinary Portland cement. The gradings of sand and coarse aggregates are given in Table 1. The same superplasticizer (a salt of naftalensulfonate) and viscosity modifying agent (a powder of welan gum) were used for all mixtures, when required. Table 1: Gradings of sand and coarse aggregate (% passing) Sieve (mm) Sand Aggregate Two series of tests were carried out in order to assess the influence of the paste volume and the W/C. In the first series, five concrete mixtures were batched using the same W/C (0.6) but with different paste volumes (30, 32, 34, 36 and 40%). In the second series, four concrete mixtures were batched using a smaller W/C (0.4) and different paste volumes (36, 38, 40 and 42%). The paste volume is defined as the fraction of the concrete volume which is not occupied by the aggregates. Paste volume thus contains cement, water and air bubbles. Similarly, the volume of mortar is obtained by adding the paste volume to the volume occupied by the sand particles. The composition of those concrete mixtures is given in Tables 2 and 3. Table 2 : Composition of concrete mixtures for the first series of tests Concrete mixture Cement Type 10 (kg/m 3 ) Water (kg/m 3 ) Sand (kg/m 3 ) Aggregate (kg/m 3 ) Air-entraining agent (ml/kgc) Paste volume (%) Mortar volume (%) The third and fourth series of tests were carried to study the influence of superplasticizer on concrete mixtures all having the same W/C (0.4). In the third series, six concrete mixtures were batched using different dosages of superplasticizer (2.8, 4.7, 5.9, 7.0, 8.2, and 9.5 ml/kg of cement) with a constant paste volume constant (36%). In the fourth series, four mixtures were batched with different dosage of superplasticizer (2.6, 3.9, 5.6, and 6.8 ml/kg of cement) Page 3

4 and a higher paste volume (40%). The composition of those concrete mixtures is given in Tables 4 and 5. Table 3 : Composition of concrete mixtures for the second series of tests Concrete Mixture Cement Type 10 (kg/m 3 ) Water (kg/m 3 ) Sand (kg/m 3 ) Aggregate (kg/m 3 ) Air-entraining agent (ml/kgc) Superplasticizer (ml/kgc) Welan gum (g/kgc) Paste volume (%) Mortar volume (%) Table 4 : Composition of concrete mixtures for the third series of tests Concrete mixture Cement Type 10 (kg/m 3 ) Water (kg/m 3 ) Sand (kg/m 3 ) Aggregate (kg/m 3 ) Air-entraining agent (ml/kgc) Superplasticizer (ml/kgc) Welan gum (g/kgc) Paste volume (%) Mortar volume (%) Table 5 : Composition of concrete mixtures for the fourth series of tests Concrete mixture Cement Type 10 (kg/m 3 ) Water (kg/m 3 ) Sand (kg/m 3 ) Aggregate (kg/m 3 ) Air entraining agent (ml/kgc) Superplasticizer SPN (ml/kgc) Welan gum (g/kgc) Paste volume (%) Mortar volume (%) The influence of the viscosity modifying agent was investigated in the fifth series of tests for concrete mixtures having the same W/C (0.4), paste volume (36%), and superplasticizer dosage (13.5 ml/kg of cement) but made with different dosages of viscosity modifying agent (1, 2 and 3 g/kg of cement). The compositions of those concrete mixtures are given in Table Concrete tests The rheological properties of fresh concrete were measured using the IBB rheometer [7]. Before testing, the rheometer bowl was first filled with concrete and placed on the apparatus. The bowl was later raised and a H-shape shaft was inserted into concrete. The test begins when the shaft is put in motion following a planetary path at an increasing speed. While the speed is later decreased, the torque applied on the shaft is measured and plotted against the Page 4

5 shear rate. According to the Bingham model, it is thus possible to calculate the yield stress, H (N-m), and plastic viscosity G (N-m-s). Table 6 : Composition of concrete mixtures for the fifth series of tests Concrete mixture Cement Type 10 (kg/m 3 ) Water (kg/m 3 ) Sand (kg/m 3 ) Aggregate (kg/m 3 ) Air-entraining agent (ml/kgc) Superplasticizer SPN (ml/kgc) Welan gum (g/kgc) Paste volume (%) Mortar volume (%) For each concrete specimen, the total surface of compaction voids was measured using a method developed in our laboratory. Numerical photographs (in 256 shades of grey) were first taken on both vertical sides of the specimen and cropped to avoid border effects (Figure 1). For homogeneity purposes, all the photographs were taken with exactly the same set-up and lightning conditions. Contrast and brightness were thus adjusted to better distinguish the compaction voids, and thresholding was further applied in order to obtain a binary image with black voids on a white background (Figure 2). After applying a median filter to remove noise from the image, the void content (V) is simply obtained by calculation the percentage of black pixels on the image. (a) original image (b) binary image Figure 1: Photograph of a concrete surface (sample V) 3. RESULTS All the test results are summarized in Table 7 where V mr represents the volume of mortar (expressed as a percentage of the volume of concrete), S the slump value (in mm), and V the surface of voids at the surface of concrete (in %). Table 7 shows that 7 out of the 24 concrete mixtures were quite stiff with slump values lower than 60 mm. Those mixtures are associated with high values of yield stress (> 18 N-m). Figure 2 shows a photograph of the stiffer concrete mixture (mixture 11) with no slump (S = 2 mm) and a very large amount of compaction voids on the concrete surface. The relationship between the plastic viscosity (H) and the surface of void content (V) is illustrated in Figure 3 for all the concrete mixtures tested. It is significant to notice that the only four concrete mixtures having a void content larger than 4% were very stiff Page 5

6 (slump < 30 mm). Otherwise, the void content seems not influenced by the viscosity of the mixture. Séries 1 Series 2 Series 3 Series 4 Series 5 Table 7 : Summary of the test results Mixture V mr (%) S (mm) H (Nm-s) G (Nm) V (%) * 25.7* * 26.8* * 34.6* 46.49** * rheological properties are not reliable because the mixture is too stiff ** number of voids is too high (cf. Figure 2) and was not taken into account in the analysis Figure 4 shows the relationship between the yield stress of concrete mixtures (G) and the void content (V). It seems to exist a threshold value of yield stress (G! 25 N-m) beyond which the void content could increase very significantly. It is very important to notice that the four stiffer mixtures (slump < 30 mm) having a void content larger than 4% are all associated with a yield stress slightly larger that the threshold value. For lower values of yield stress (G < 25 N-m), there is no clear influence of the yield stress on the void content. Page 6

7 Series 1 Series 2 Series 3 Series 4 Series 5 V% 6 4 Slump < 30 mm H (Nm s) Figure 2 : Photograph of stiff concrete mixture (mixture 11) having a lot of compaction voids on surface Figure 3 : Influence of plastic viscosity on surface void content %V Slump < 30 mm Series 1 Series 2 Series 3 Series 4 Series 5 %V Series 2 (W/C=0.4) Series 1 (W/C=0.6) G (Nm) mortar volume % Figure 4 : Influence of yield stress on void content Figure 5 : Influence of mortar volume on void content Table 7 indicates that, for the first two series of mixtures, the slump (S) and apparent yield stress (G) decrease with an increase of the mortar volume (or paste volume since the volume of sand is kept constant) while the torque plastic viscosity (H) seems very little influenced. It was also found that the void content decreases with an increase of the mortar volume, as illustrated in Figure 5. For low mortar volumes, the void content decreases abruptly with the increase of the mortar volume, but it rapidly tends to an asymptotic value. Figure 5 also points out the paramount importance of the W/C. A concrete mixture with a lower W/C ratio requires a higher volume of mortar to avoid reach the asymptote and limit the void content on the concrete surface. For mixtures of series 3 and 4, the only differences between mixtures are the superplasticizer dosage (2.8 to 9.5 ml/kg of cement), and the paste volume (36% for series 3 and 40% for series 4). For those mixtures, Table 7 indicates that a higher dosage of Page 7

8 superplasticizer significantly increases the slump (S) and reduces the yield stress (G). It also increases the plastic viscosity (H) but in a much lesser extent. Table 7 also indicates that the concrete mixtures with a higher paste content have a lower yield stress and a lower viscosity. Although those mixtures cover a wide range of rheological properties, no significant differences were found as regards with the void content except for very stiff mixtures (slump < 30 mm) which are associated to high void contents. Mixtures of series 5 were similar to those of series 3, except that a viscosity-modifying agent was added at different dosages. Table 7 indicates that the adding of this admixture has significantly increased the viscosity of concrete. The concrete mixtures of series 5 generally have a higher void content than those of series 3, which suggests that the presence of the viscosity-modifying admixture could slightly increase the void content. 4. DISCUSSION The test results clearly indicate that any increase in paste or mortar volume reduces the yield stress of concrete. Those results are consistent with previous findings by de Larrard [9] showing that the yield stress is governed by the contribution of the solid phase of the concrete mixture, while the plastic viscosity is rather governed by the liquid phase as regards with the constraint movement of particles into fresh concrete. When the volume of paste is increased, the aggregates are distant from one another, the friction between them decreases, and the yield stress is reduced. Test results also confirm the findings of many previous studies showing that superplasticizers reduce very significantly the yield stress of concrete mixtures. However, the test results also indicate that the superplasticizer also increased the plastic viscosity of concrete mixture despite the fact that superplasticizers are known to have little or no influence on the viscosity of concrete [10,11]. The authors believe that the superplasticizer used in this study contains a small amount of viscosity-modifying agent (most probably to prevent segregation in job site conditions), which could explain the results obtained. The 24 concrete mixtures tested in this study cover a wide range of rheological properties. It is clear that the viscosity does not influence the void content even though it ranged from about 3 to 21 N-m-s. It was observed that void content becomes really significant (V > 4%) only when the yield value exceeds 25 N-m and the mixture was very stiff (slump < 30 mm) 1. For all other mixtures, there is no clear relationship between the yield stress and the void content. Those results suggest that the rheological properties exert only a little influence on the occurrence of compaction voids at the surface of concrete. To better understand the mechanism of void formation, a freshly-mixed concrete of normal consistency (slump! 100 mm) was cast into a mould made with a transparent plexiglass sidewall. The concrete mixture was made using a high volume of coarse aggregate and a relatively low volume of cement paste and mortar. After placing, many coarse aggregates were seen against the wall with a lot of large voids between them. A soon as the vibration begins, the those voids disappeared in a few seconds due to a denser rearrangement of the granular skeleton and the coming of paste and mortar from inside the mixture to the sidewall. 1 For such stiff mixtures, the workability was so bad that they would certainly be rejected at the job site. Page 8

9 During vibration, the aggregates swing at a high frequency, which almost eliminate the friction between aggregate and, consequently, reduces the yield stress of the concrete mixture to a near-zero value [8]. The influence of vibration thus overwhelm the influence of the rheological properties. This explains why no relationship was found between the rheological properties and the void content of the tested concrete mixture. However, if a concrete mixture contains a higher mortar volume, the granular skeleton will be less closely packed and the mortar fraction would more easily pass through the aggregates to reach the sidewall of the mould. A larger amount a mortar is also available to fill the empty spaces near the wall. Those two effects both contribute to reduce the void content. However, if the mortar volume is sufficient to allow a good filling of the voids initially left near the sidewall, any further increase in the mortar volume will not reduces the void content which explain the asymptote observed in Figure 5. This Figure also indicates very clearly that the mortar volume is the key parameter as regards with the void content and that its influence is much more important than the rheological properties of the concrete mixture. 5. CONCLUSION The results of this experimental study show that the extent of compaction voids on concrete surfaces is mainly related to the volume of mortar in the concrete mixture. The results also indicate that, except for very stiff mixture having a very slow slump value (S < 30 mm) and high yield stress (G > 25 N-m) which are responsible for high void content (V > 4%), the rheological properties of concrete only have a little influence on the amount of compaction voids at the surface of concrete. The results also strongly suggest that the vibration is the most important parameter as regards with the occurrence of compaction voids. From a practical point of view, it means that the better way to avoid compaction voids on concrete surfaces is to select adequate mixture proportions (with a sufficiently high volume of paste and mortar), and to carefully control the quality of vibration during placing of concrete at job site. In those conditions, the rheological properties of concrete are of little influence. It is likely; however, that the rheological properties, and most especially the yield stress, may influence the void content for an improper mixture design or a deficient vibration technique. It is also possible that the rheological properties exert an influence on the void content of selfcompacting concretes, which are put in place without using vibration. Further research is needed to confirm those hypothesis. REFERENCES [1] Lallemant I., Hétérogénéités de teinte des parements en béton : caractérisation et identification des mécanismes, Ph.D. Thesis, Université de Cergy-Pontoise, (2001). [2] Lemaire G., Contribution à la maîtrise de la qualité des parements en béton, Ph.D. Theses, Université Paul Sabatier, Toulouse, (2003). [3] Monks W. L., An investigation into the Incidence of Colour Variation in Formed Concrete Surfaces, Technical report, Cement and Concrete Association, (1974). [4] Murphy W.E., The influence of Concrete Mix Proportions and Type of Form Face on the Appearance of Concrete, Technical report, Cement and Concrete Association, (1967). [5] Kinnear R.G., Concrete Surface Blemishes, a Classification of Surface Defects and Some Influences of Formwork linings, release agents and concrete pressure on the appearance of concrete finishes, Cement and Concrete Association, (1964). [6] Buzzi S., The Cement in Exposed Concrete Technology, Il Cemento (4) (1986) Page 9

10 [7] Beaupré Denis, Rheology of high-performance Shotcrete, Ph.D. Thesis, University of British Columbia, [8] Tattersall, G.H. & Banfill P.F.G., The Rheology of Fresh Concrete, (London, Pitman, 1983). [9] De Larrard F., Structures granulaires et formulation des bétons, (études et recherches des laboratoires des ponts et chaussées OA 34, 2000) [10] Ramachandran V. S., Malhotra V.M., Jolicoeur C. and Spiratos N., Superplasticizers : Properties and Applications in Concrete, (Canmet, 1998). [11] Chapdelaine F., Étude sur la rhéologie du béton frais, mémoire de maîtrise, Université Laval, (1998). Page 10